Your cookie choices

We use necessary cookies to make this site work. With your permission, analytics cookies help us understand usage and marketing cookies support advertising.

You can reject non-essential cookies and still use the site. Change your choices anytime in Cookie settings.

Cookie settingsCookie policy

Perplexity (Sonar API) review

Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026

We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings. Partners with a live offer in your country come first, then every other brand by score.

Our verdict

Perplexity's Sonar API suits teams that want a single vendor's answer models without shopping across a general purpose catalogue from another provider, and the OpenAI-compatible endpoint keeps an existing client in place. Input fees run $1 to $3 and output fees $1 to $15, with a largest context window of 200,000.

Key facts

Best for
A single vendor's answer models on an OpenAI-compatible API
Input price from
$1 to $3↗
Output price from
$1 to $15↗
Models offered
Sonar, Sonar Pro, Sonar Reasoning Pro, Sonar Deep Research↗
Largest context window
200,000↗
OpenAI-compatible API
Yes↗
Legal entity
Perplexity AI, Inc.↗
Headquarters
United States↗
Founded
2022↗
Official website
perplexity.ai

Facts checked September 26, 2026, 5 sources

Apps and profiles

Elsewhere
Wikidata

Pros and cons

Pros

  • A focused catalogue rather than a long list: Sonar, Sonar Pro and Sonar Reasoning Pro.
  • OpenAI-compatible API, so an existing client can stay with a changed base URL.
  • A 200,000 context window on Sonar Pro.

Cons

  • Input fees of $1 to $3 sit above the site median of €0.07.
  • Output fees of $1 to $15 sit above the site median of €0.23.
  • A 200,000 context window falls short of the site median of 1,000,000.

Ready to try Perplexity (Sonar API)?

Scored 8.5 / 10 by DeployCue.

See alternatives

Who it suits

  • Teams that want a single vendor's answer models on the endpoint rather than a router across many providers.
  • Developers already on an OpenAI-compatible client, since this API is OpenAI-compatible.
  • Builders whose documents fit a 200,000 context window and whose budget absorbs input fees of $1 to $3.

Who should look elsewhere

  • Azure AI Foundry, if you need image, audio or coding models beside a chat model.
  • Salad, if token cost decides it: input price from $0.06 is below the $1 to $3 Sonar band.
  • Supascale AI, if you want a gateway across many model families rather than a single vendor's own line.

Where it sits among rivals

Perplexity ranks 66 of 71 on input price from, 59 of 70 on output price from and 58 of 72 on largest context window. None of those rankings sits near the front, so a buyer opening this group on cost has other rows to read, and a buyer who needs a very large context window has the same story.

Who runs it

The API is published by Perplexity AI, Inc., headquartered in the United States and founded in 2022. A single US entity sits behind the Sonar models, and no reseller or partner layer is named, so a buyer signing up deals with the same legal name that carries the pricing. The contract and the model catalogue stay under the same name.

How Perplexity (Sonar API) compares

Frequently asked questions

What does it cost to run answers through Sonar?+

Sonar is billed per token. Input fees run $1 to $3 and output fees $1 to $15, so long generated answers cost more than long prompts. Those figures sit above the site medians of €0.07 on input and €0.23 on output, which is worth checking before a build goes live.

Which models can I call?+

The catalogue is Sonar, Sonar Pro and Sonar Reasoning Pro. That covers a standard answer model, a pro tier, a reasoning variant and a deep research model, so a team can trade up on capability without moving provider. Only the Sonar family is served, which narrows choice if a workload needs an image, audio or open weight model.

How much context fits in a request?+

The largest context window listed is 200,000, recorded against Sonar Pro. That handles long documents, though it ranks 58 of 72 on this site and sits below the site median of 1,000,000, so very large documents are not the strong suit here.

Who runs the API and where are they based?+

The API is published by Perplexity AI, Inc., headquartered in the United States and founded in 2022. A single company carries the models, the pricing and the contract, so there is no reseller layer to work through and no second entity named on the paperwork.

Can I keep my existing client?+

Yes. The API is OpenAI-compatible, and the model strings you send are Sonar, Sonar Pro and Sonar Reasoning Pro, so an existing client can point at a different base URL rather than be rebuilt. Request and response shapes still need a check, and each name in that list maps to a different price band.

Where do I get more choice of models?+

Azure AI Foundry publishes a catalogue, gpt 6 astra, gpt 6 luna and gpt 6 sol, across general, coding, image and audio models. Sonar offers a single vendor's line instead, with no other family to compare on a single bill.

Sources

Compiled from 5 source pages, checked September 26, 2026. We did not open an account.

Perplexity (Sonar API)

A single vendor's answer models on an OpenAI-compatible API

8.5/10See alternatives